Weekly Exercises #4

library(tidyverse)
library(lubridate)   
library(openintro)   
library(maps)
library(ggmap)    
library(gplots)    
library(RColorBrewer)  
library(sf)          
library(leaflet)
library(carData)       
library(ggthemes)     
theme_set(theme_minimal())

# Starbucks locations
Starbucks <- read_csv("https://www.macalester.edu/~ajohns24/Data/Starbucks.csv")

starbucks_us_by_state <- Starbucks %>% 
  filter(Country == "US") %>% 
  count(`State/Province`) %>% 
  mutate(state_name = str_to_lower(abbr2state(`State/Province`))) 

# Lisa's favorite St. Paul places - example for you to create your own data
favorite_stp_by_lisa <- tibble(
  place = c("Home", "Macalester College", "Adams Spanish Immersion", 
            "Spirit Gymnastics", "Bama & Bapa", "Now Bikes",
            "Dance Spectrum", "Pizza Luce", "Brunson's"),
  long = c(-93.1405743, -93.1712321, -93.1451796, 
           -93.1650563, -93.1542883, -93.1696608, 
           -93.1393172, -93.1524256, -93.0753863),
  lat = c(44.950576, 44.9378965, 44.9237914,
          44.9654609, 44.9295072, 44.9436813, 
          44.9399922, 44.9468848, 44.9700727)
  )

#COVID-19 data from the New York Times
covid19 <- read_csv("https://raw.githubusercontent.com/nytimes/covid-19-data/master/us-states.csv")

Put your homework on GitHub!

If you were not able to get set up on GitHub last week, go here and get set up first. Then, do the following (if you get stuck on a step, don’t worry, I will help! You can always get started on the homework and we can figure out the GitHub piece later):

• Create a repository on GitHub, giving it a nice name so you know it is for the 4th weekly exercise assignment (follow the instructions in the document/video).
  
• Copy the repo name so you can clone it to your computer. In R Studio, go to file –> New project –> Version control –> Git and follow the instructions from the document/video.
  
• Download the code from this document and save it in the repository folder/project on your computer.
  
• In R Studio, you should then see the .Rmd file in the upper right corner in the Git tab (along with the .Rproj file and probably .gitignore).
  
• Check all the boxes of the files in the Git tab under Stage and choose commit.
  
• In the commit window, write a commit message, something like “Initial upload” would be appropriate, and commit the files.
  
• Either click the green up arrow in the commit window or close the commit window and click the green up arrow in the Git tab to push your changes to GitHub.
  
• Refresh your GitHub page (online) and make sure the new documents have been pushed out.
  
• Back in R Studio, knit the .Rmd file. When you do that, you should have two (as long as you didn’t make any changes to the .Rmd file, in which case you might have three) files show up in the Git tab - an .html file and an .md file. The .md file is something we haven’t seen before and is here because I included keep_md: TRUE in the YAML heading. The .md file is a markdown (NOT R Markdown) file that is an interim step to creating the html file. They are displayed fairly nicely in GitHub, so we want to keep it and look at it there. Click the boxes next to these two files, commit changes (remember to include a commit message), and push them (green up arrow).
  
• As you work through your homework, save and commit often, push changes occasionally (maybe after you feel finished with an exercise?), and go check to see what the .md file looks like on GitHub.
  
• If you have issues, let me know! This is new to many of you and may not be intuitive at first. But, I promise, you’ll get the hang of it!

Instructions

• Put your name at the top of the document.

• For ALL graphs, you should include appropriate labels.

• Feel free to change the default theme, which I currently have set to theme_minimal().

• Use good coding practice. Read the short sections on good code with pipes and ggplot2. This is part of your grade!

• When you are finished with ALL the exercises, uncomment the options at the top so your document looks nicer. Don’t do it before then, or else you might miss some important warnings and messages.

Warm-up exercises from tutorial

These exercises will reiterate what you learned in the “Mapping data with R” tutorial. If you haven’t gone through the tutorial yet, you should do that first.

Starbucks locations (ggmap)

1. Add the Starbucks locations to a world map. Add an aesthetic to the world map that sets the color of the points according to the ownership type. What, if anything, can you deduce from this visualization?

world <- get_stamenmap(
    bbox = c(left = -180, bottom = -57, right = 179, top = 82.1), 
    maptype = "terrain",
    zoom = 2)

ggmap(world) +
  geom_point(data = Starbucks, 
             aes(x = Longitude, y = Latitude, color = `Ownership Type`), 
             alpha = .3, 
             size = .2) +
  theme_map() +
  theme(legend.background = element_blank())

There are less Starbucks that are franchised and joint ventures; most of them are company owned or are licensed. Besides, Starbucks that are joint ventures seem only exist in Asia and Europe.

2. Construct a new map of Starbucks locations in the Twin Cities metro area (approximately the 5 county metro area).

Twin_cities_metro <- get_stamenmap(
    bbox = c(left = -93.4140, bottom = 44.8705, right = -92.9574, top = 45.0803), 
    maptype = "terrain",
    zoom = 12)

ggmap(Twin_cities_metro) +
  geom_point(data = Starbucks, 
             aes(x = Longitude, y = Latitude), 
             alpha = .3, 
             size = 1.5,
             color = "darkred") +
  theme_map() +
  theme(legend.background = element_blank())

3. In the Twin Cities plot, play with the zoom number. What does it do? (just describe what it does - don’t actually include more than one map).
   With a smaller zoom number, there will be less detail information about roads shown in the plot.

4. Try a couple different map types (see get_stamenmap() in help and look at maptype). Include a map with one of the other map types.

Twin_cities_metro <- get_stamenmap(
    bbox = c(left = -93.4140, bottom = 44.8705, right = -92.9574, top = 45.0803), 
    maptype = "watercolor",
    zoom = 12)

ggmap(Twin_cities_metro) +
  geom_point(data = Starbucks, 
             aes(x = Longitude, y = Latitude), 
             alpha = .3, 
             size = 1.7) +
  theme_map() +
  theme(legend.background = element_blank())

5. Add a point to the map that indicates Macalester College and label it appropriately. There are many ways you can do think, but I think it’s easiest with the annotate() function (see ggplot2 cheatsheet).

Twin_cities_metro <- get_stamenmap(
    bbox = c(left = -93.4140, bottom = 44.8705, right = -92.9574, top = 45.0803), 
    maptype = "watercolor",
    zoom = 12)

ggmap(Twin_cities_metro) +
  geom_point(data = Starbucks, 
             aes(x = Longitude, y = Latitude), 
             alpha = .3, 
             size = 1.7) +
  annotate("point", x = -93.1704, y = 44.9401, color = "midnightblue", size = 3, alpha = 1) +
  annotate("text", x = -93.1704, y = 44.935, label = "Mac", color = "steelblue4") +
  theme_map() +
  theme(legend.background = element_blank())

Choropleth maps with Starbucks data (geom_map())

The example I showed in the tutorial did not account for population of each state in the map. In the code below, a new variable is created, starbucks_per_10000, that gives the number of Starbucks per 10,000 people. It is in the starbucks_with_2018_pop_est dataset.

census_pop_est_2018 <- read_csv("https://www.dropbox.com/s/6txwv3b4ng7pepe/us_census_2018_state_pop_est.csv?dl=1") %>% 
  separate(state, into = c("dot","state"), extra = "merge") %>% 
  select(-dot) %>% 
  mutate(state = str_to_lower(state))

starbucks_with_2018_pop_est <-
  starbucks_us_by_state %>% 
  left_join(census_pop_est_2018,
            by = c("state_name" = "state")) %>% 
  mutate(starbucks_per_10000 = (n/est_pop_2018)*10000)

6. dplyr review: Look through the code above and describe what each line of code does.
   It first read a dataset from the dropbox, seperated the state variable into two new variables: dot, whose values are nothing, and state, which contains only the state names, and then selected only the new state variable. After that, by using the mutate() function, all state names are transformed to lower case. Finally, the transformed dataset is saved and named as census_pop_est_2018.
   Then a new dataset named starbucks_with_2018_pop_est is created by left joining starbucks_us_by_state with census_pop_est_2018 just created, and a new variable is created that gives the number of Starbucks per 10,000 people.

7. Create a choropleth map that shows the number of Starbucks per 10,000 people on a map of the US. Use a new fill color, add points for all Starbucks in the US (except Hawaii and Alaska), add an informative title for the plot, and include a caption that says who created the plot (you!). Make a conclusion about what you observe.

US_map <- map_data("state")

starbucks_with_2018_pop_est %>%
  ggplot() +
  geom_map(map = US_map,
           aes(map_id = state_name, fill = starbucks_per_10000)) +
  geom_point(data = Starbucks %>% filter(Country == "US", `State/Province` != "AK" & `State/Province` != "HI"),
             aes(x = Longitude, y = Latitude),
                 size = .25,
                 alpha = .1,
                 color = "goldenrod") +
  expand_limits(x = US_map$long, y = US_map$lat) +
  labs(title = "Number of Starbucks per 10,000 people in US",
       subtitle = "With all Starbucks plotted as points",
       fill = "",
       caption = "Plotted by Xintan Xia") +
  scale_fill_viridis_c(option = "magma") +
  theme_map()

It’s shown clearly that the for some states along the east coast, where numerous Starbucks cluster, the density of Starbucks per 10,000 people is actually quite low, about 0.5-0.75 Starbucks per 10,000 people. Starbucks also tend to cluster in California, and it’s obvious seen from the plot that there are less number of Starbucks in Washington than in California, yet it’s also clearly shown that the Starbucks density per 10,000 people in Washington is higher than that in California.
In conclusion, the total number of Starbucks and the density of Starbucks per 10,000 people are two different stories.

A few of your favorite things (leaflet)

8. In this exercise, you are going to create a single map of some of your favorite places! The end result will be one map that satisfies the criteria below.

• Create a data set using the tibble() function that has 10-15 rows of your favorite places. The columns will be the name of the location, the latitude, the longitude, and a column that indicates if it is in your top 3 favorite locations or not. For an example of how to use tibble(), look at the favorite_stp_by_lisa I created in the data R code chunk at the beginning.

• Create a leaflet map that uses circles to indicate your favorite places. Label them with the name of the place. Choose the base map you like best. Color your 3 favorite places differently than the ones that are not in your top 3 (HINT: colorFactor()). Add a legend that explains what the colors mean.

• Connect all your locations together with a line in a meaningful way (you may need to order them differently in the original data).

• If there are other variables you want to add that could enhance your plot, do that now.

favorite_stp_by_Xia <- tibble(
  place = c("Home", "Macalester College", "Spirit1919", 
            "Long Museum", "Previous home", "Qinhan Noodle shop",
            "Haidilao Hot pot", "Jianhui restaurant", "Girlfriend's University",
            "My Middle School"),
  long = c(108.89440, -93.1712321, 108.88881, 
           121.56973, 108.9053, 108.8284, 
           108.8742, 108.9463, 87.303,
           108.9009),
  lat = c(34.21168, 44.9378965, 34.20908,
          31.21816, 34.2224, 34.2356, 
          34.2113, 34.2140, 44.0178,
          34.2282),
  top_3 = c("Yes", "Yes", "No",
            "No", "Yes", "No",
            "No", "No", "No",
            "No")
  )

pal <- colorFactor(palette = "viridis",
                   domain = favorite_stp_by_Xia$top_3)

leaflet(favorite_stp_by_Xia) %>%
  addProviderTiles(providers$Stamen.Watercolor) %>%
  addCircles(color = ~pal(top_3),
             label = ~place) %>%
  addPolylines(lng = ~long,
               lat = ~lat,
               color = col2hex("hotpink")) %>%
  addLegend("bottomright",
            pal = pal,
            values = ~top_3,
            title = "Top 3 places?")

Revisiting old datasets

This section will revisit some datasets we have used previously and bring in a mapping component.

Bicycle-Use Patterns

The data come from Washington, DC and cover the last quarter of 2014.

Two data tables are available:

• Trips contains records of individual rentals
• Stations gives the locations of the bike rental stations

Here is the code to read in the data. We do this a little differently than usualy, which is why it is included here rather than at the top of this file. To avoid repeatedly re-reading the files, start the data import chunk with {r cache = TRUE} rather than the usual {r}. This code reads in the large dataset right away.

data_site <- 
  "https://www.macalester.edu/~dshuman1/data/112/2014-Q4-Trips-History-Data.rds" 
Trips <- readRDS(gzcon(url(data_site)))
Stations<-read_csv("http://www.macalester.edu/~dshuman1/data/112/DC-Stations.csv")

9. Use the latitude and longitude variables in Stations to make a visualization of the total number of departures from each station in the Trips data. Use either color or size to show the variation in number of departures. This time, plot the points on top of a map. Use any of the mapping tools you’d like.

Trips1 <- Trips %>% 
  select(sstation, client) %>% 
  left_join(Stations %>% select(name:long),
            by = c("sstation" = "name"))

washdc_map <- get_stamenmap(bbox = c(left = -77.17, bottom = 38.7889, right = -76.9319, top = 39.1),
                            maptype = "terrain",
                            zoom = 11)

ggmap(washdc_map) +
  geom_point(data = Trips1,
             aes(x = long, y = lat),
             size = .1,
             alpha = .1) +
  theme_map()

10. Only 14.4% of the trips in our data are carried out by casual users. Create a plot that shows which area(s) have stations with a much higher percentage of departures by casual users. What patterns do you notice? Also plot this on top of a map. I think it will be more clear what the patterns are.

Trips2 <- Trips1 %>%
  group_by(sstation) %>%
  mutate(is_casual = client == "Casual", prop_casual = sum(is_casual)/n()) 
 

ggmap(washdc_map) +
  geom_point(data = Trips2,
             aes(x = long, y = lat, color = prop_casual),
             size = .2) +
  scale_color_viridis_c(option = "inferno") +
  labs(title = "Proportion of casual users for stations") +
  theme_map() +
  theme(legend.title = element_blank())

We can see from the plot that those stations with a much higher proportion of casual users are mostly around places where all the monuments are, which should be visited quite often by tourists.

COVID-19 data

The following exercises will use the COVID-19 data from the NYT.

11. Create a map that colors the states by the most recent cumulative number of COVID-19 cases (remember, these data report cumulative numbers so you don’t need to compute that). Describe what you see. What is the problem with this map?

covid_most_recent <- covid19 %>%
  filter(date == max(date)) %>%
  mutate(state = str_to_lower(state))

states_map <- map_data("state")

covid_most_recent %>%
  ggplot() +
  geom_map(map = states_map,
           aes(map_id = state, fill = cases)) +
  expand_limits(x = states_map$long, y = states_map$lat) +
  labs(title = "Most recent cumulative COVID-19 case count in US",
       fill = "Case Count") +
  scale_fill_viridis_c(option = "viridis", label = scales::comma) +
  theme_map()

The plot implies that all other states are currently in almost the same level, except California, Texas, Florida, Illinois, and New York, in which Texas and California have the largest number of cumulative case count currently.
The question is that this graph can’t give people a clear comparison between the conditions of the majority of states, which may be caused by the continuous color scale with quite a large interval between each scales. The fact that this plot doesn’t take density into consideration also attributes to the problem. Right now, the total case counts are plotted with respect to state area, but the pure cumulative case counts are not that much informative, under this circumstance, and large areas can be misleading as well. In reflecting the seriousness of current conditions, we should consider the density.

12. Now add the population of each state to the dataset and color the states by most recent cumulative cases/10,000 people. See the code for doing this with the Starbucks data. You will need to make some modifications.

census_pop_est_2018 <- read_csv("https://www.dropbox.com/s/6txwv3b4ng7pepe/us_census_2018_state_pop_est.csv?dl=1") %>% 
  separate(state, into = c("dot","state"), extra = "merge") %>% 
  select(-dot) %>% 
  mutate(state = str_to_lower(state))

covid_most_recent %>%
  left_join(census_pop_est_2018,
            by = "state") %>%
  mutate(cases_per_10000 = (cases/est_pop_2018)*10000) %>%
  ggplot() +
  geom_map(map = states_map,
           aes(map_id = state, fill = cases_per_10000)) +
  scale_fill_viridis_c(option = "viridis") +
  expand_limits(x = states_map$long, y = states_map$lat) +
  labs(title = "Most recent COVID-19 case counts in US",
       subtitle = "Cumulative case counts per 10,000 people",
       fill = "") +
  theme_map()

13. CHALLENGE Choose 4 dates spread over the time period of the data and create the same map as in exercise 12 for each of the dates. Display the four graphs together using faceting. What do you notice?

covid_four_dates <- covid19 %>%
  filter(date == as.Date("2020-3-28") | date == as.Date("2020-8-28") | date == as.Date("2021-1-28") | date == max(date)) %>%
  mutate(state = str_to_lower(state))
  

covid_four_dates %>%
  left_join(census_pop_est_2018,
            by = "state") %>%
  mutate(cases_per_10000 = (cases/est_pop_2018)*10000) %>%
  arrange(state, date) %>%
  ggplot() +
  geom_map(map = states_map,
           aes(map_id = state, fill = cases_per_10000)) +
  scale_fill_viridis_c(option = "A") +
  expand_limits(x = states_map$long, y = states_map$lat) +
  facet_wrap(vars(date)) +
  labs(title = "Most recent COVID-19 case counts in US",
       subtitle = "Cumulative case counts per 10,000 people",
       fill = "") +
  theme_map() +
  theme(legend.background = element_blank(),
        legend.position = "bottom")

The specific pattern (the relative conditions between states) observed in 2020-8-28 seems not to change much until the most recent day.

Minneapolis police stops

These exercises use the datasets MplsStops and MplsDemo from the carData library. Search for them in Help to find out more information.

14. Use the MplsStops dataset to find out how many stops there were for each neighborhood and the proportion of stops that were for a suspicious vehicle or person. Sort the results from most to least number of stops. Save this as a dataset called mpls_suspicious and display the table.

mpls_suspicious <- MplsStops %>%
  mutate(is_suspicious = problem == "suspicious") %>%
  group_by(neighborhood) %>%
  summarize(num_stops = n(), prop_suspicious = sum(is_suspicious)/num_stops) %>%
  arrange(desc(num_stops))

mpls_suspicious

15. Use a leaflet map and the MplsStops dataset to display each of the stops on a map as a small point. Color the points differently depending on whether they were for suspicious vehicle/person or a traffic stop (the problem variable). HINTS: use addCircleMarkers, set stroke = FAlSE, use colorFactor() to create a palette.

Mstops_pal <- colorFactor(palette = "viridis",
                          domain = MplsStops$problem)

leaflet(data = MplsStops) %>%
  addTiles() %>%
  addCircleMarkers(color = ~Mstops_pal(problem),
                   stroke = FALSE,
                   radius = 1.5,
                   opacity = .1) %>%
  addLegend(pal = Mstops_pal,
            values = ~problem,
            position = "bottomright",
            title = "")

16. Save the folder from moodle called Minneapolis_Neighborhoods into your project/repository folder for this assignment. Make sure the folder is called Minneapolis_Neighborhoods. Use the code below to read in the data and make sure to delete the eval=FALSE. Although it looks like it only links to the .sph file, you need the entire folder of files to create the mpls_nbhd data set. These data contain information about the geometries of the Minneapolis neighborhoods. Using the mpls_nbhd dataset as the base file, join the mpls_suspicious and MplsDemo datasets to it by neighborhood (careful, they are named different things in the different files). Call this new dataset mpls_all.

mpls_nbhd <- st_read("Minneapolis_Neighborhoods/Minneapolis_Neighborhoods.shp", quiet = TRUE)

mpls_all <- mpls_nbhd %>% 
  left_join(mpls_suspicious,
            by = c("BDNAME" = "neighborhood")) %>%
  left_join(MplsDemo,
                by = c("BDNAME" = "neighborhood"))

17. Use leaflet to create a map from the mpls_all data that colors the neighborhoods by prop_suspicious. Display the neighborhood name as you scroll over it. Describe what you observe in the map.

mplsall_pal = colorNumeric(palette = "inferno",
                           domain = mpls_all$prop_suspicious)

leaflet(data = mpls_all) %>%
  addTiles() %>%
  addPolygons(stroke = FALSE,
              fillColor = ~mplsall_pal(prop_suspicious),
              label = ~BDNAME,
              fillOpacity = .7) %>%
  addLegend(position = "bottomright",
            pal = mplsall_pal,
            values = ~prop_suspicious)

The proportion varies much across neighborhoods, and the distinction between neighborhoods with different proportions is quite clear. The general trend is that neighborhoods with a high proportions of stops that were for a suspicious vehicle or person seem to cluster in the south, while those with low proportions are mostly in the north. Yet there are also some neighborhoods in the south with a low proportion, such as Tangletown, Kenwood, etc.

18. Use leaflet to create a map of your own choosing. Come up with a question you want to try to answer and use the map to help answer that question. Describe what your map shows.

mplsall_pal = colorNumeric(palette = "inferno",
                           domain = mpls_all$prop_white)

leaflet(data = mpls_all) %>%
  addTiles() %>%
  addPolygons(stroke = FALSE,
              fillColor = ~mplsall_pal(white),
              popup = ~paste(BDNAME,": ",
                            white,
                            sep=""),
              fillOpacity = .7) %>%
  addLegend(position = "bottomright",
            pal = mplsall_pal,
            values = ~white,
            title = "White proportion")

I’m trying to use my map to answer the question regarding the proportion of white people in Minneapolis neighborhoods.
With a few neighborhoods with unknown proportion of white residents, it can be seen from the map that neighborhoods are mostly with a high proportion of white, except those right in the center and in the northwest. It’s also interesting to see that instead of changing abruptly from above 0.75, the proportion of white actually decreases gradually from the outer communities to the inner ones, where people with other races are dominant residents.

GitHub link

19. Below, provide a link to your GitHub page with this set of Weekly Exercises. Specifically, if the name of the file is 04_exercises.Rmd, provide a link to the 04_exercises.md file, which is the one that will be most readable on GitHub.
    Here’s the link.

DID YOU REMEMBER TO UNCOMMENT THE OPTIONS AT THE TOP?